External Validation and Update of a Prediction Rule for the Duration of Sickness Absence Due to Common Mental Disorders.

Only 220 of 596 employees who reported sick agreed to participate, so the analysed group may not represent all employees with common mental disorder absence. The sample had only 7% teachers, unlike the development sample with 54% teachers, so the rule may not transfer well. The study participants may have had more severe common mental disorders and longer absence than all employees who reported sick in the same period. Only 18 employees were still absent at 12 months, so the rule could not be validated at 12 months. Work-related factors were self-reported, so mood or expectations could affect them. The updated rule was fitted to the same sample, so its performance in other workplaces is not yet known. The rule still needs a simpler format and cut-off points before practical use. Companies with sickness absence insurance are generally small companies with up to 100 employees.

Do not treat the updated rule as ready for referral decisions. The AUC 0.74 came from recalibrating and adding variety in work in the same 211 participants, and the authors say it still needs testing in other validation studies.